def parse(self, raw):
global tmp_content
global tmp_alignment
global tmp_grid_col_to
global tmp_grid_col_from
global tmp_grid_row_to
global tmp_grid_row_from
global tmp_vertical_align
# At first, content is blank
tmp_content = ""
# because there's no line
tmp_objs = []
# Get <cell> node attributes
tmp_alignment = re.search(r"alignment=\"(\w+)\"", raw).group(1)
tmp_grid_row_from = re.search(r"gridRowFrom=\"(\d+)\"", raw).group(1)
tmp_grid_row_to = re.search(r"gridRowTill=\"(\d+)\"", raw).group(1)
tmp_grid_col_from = re.search(r"gridColFrom=\"(\d+)\"", raw).group(1)
tmp_grid_col_to = re.search(r"gridColTill=\"(\d+)\"", raw).group(1)
# TODO: don't understand, attribute with value = 0 will be returned as undef by twig
tmp_grid_row_from = tmp_grid_row_from if tmp_grid_row_from else 0
tmp_grid_row_to = tmp_grid_row_to if tmp_grid_row_to else 0
tmp_grid_col_from = tmp_grid_col_from if tmp_grid_col_from else 0
tmp_grid_col_to = tmp_grid_col_to if tmp_grid_col_to else 0
tmp_vertical_align = re.search(r"verticalAlignment=\"(\w+)\"", raw).group(1)
tmpRaw = raw
tmpRaw = re.sub(r"^<cell[^>]*>\n", "", tmpRaw)
tmpRaw = re.sub(r"</cell>(\n)?$", "", tmpRaw)
tmpRaw = re.sub(r"<leftBorder[^>]*>\n", "", tmpRaw, 1)
tmpRaw = re.sub(r"<rightBorder[^>]*>\n", "", tmpRaw, 1)
tmpRaw = re.sub(r"<topBorder[^>]*>\n", "", tmpRaw, 1)
tmpRaw = re.sub(r"<bottomBorder[^>]*>\n", "", tmpRaw, 1)
sys.path.append("./LSRRTF/")
import rtfPara
while True:
if re.search(r"^<para", tmpRaw) != None:
para = rtfPara.rtfPara()
indexParaEnd = tmpRaw.index("</para>") + len("</para>")
tmpParaRaw = tmpRaw[:indexParaEnd]
tmpRaw = tmpParaRaw[indexParaEnd+1:]
# Set raw content
para.set_raw(tmpParaRaw)
# Update paragraph list
tmp_objs.append(para)
# Update content
tmp_content += para.get_content() + "\n"
else:
break
def parse(self, raw):
# At first, content is blank
tmp_content = ""
# because there's no object
tmp_objs = []
# Get <column> node attributes
tmp_bottom = re.search(r"b=\"(\d+)\"", raw).group(1)
tmp_top = re.search(r"t=\"(\d+)\"", raw).group(1)
tmp_left = re.search(r"l=\"(\d+)\"", raw).group(1)
tmp_right = re.search(r"r=\"(\d+)\"", raw).group(1)
tmpRaw = raw
tmpRaw = re.sub(r"^<column[^>]*>\n", "", tmpRaw)
tmpRaw = re.sub(r"</column>(\n)?$", "", tmpRaw)
sys.path.append("./LSRRTF/")
import rtfPara
import rtfTable
while True:
if re.search(r"^<para", tmpRaw) != None:
para = rtfPara.rtfPara()
indexParaEnd = tmpRaw.index("</para>") + len("</para>")
tmpParaRaw = tmpRaw[:indexParaEnd]
tmpRaw = tmpRaw[indexParaEnd + 1 :]
# Set raw content
para.set_raw(tmpParaRaw)
# Update paragraph list
tmp_objs.append(para)
# Update content
tmp_content = tmp_content + para.get_content() + "\n"
elif re.search(r"^<dd", tmpRaw) != None:
sys.strerr.write("Para::dd::Neimplementovano\n")
indexDdEnd = tmpRaw.index("</dd>") + len("</dd>")
# tmpDdRaw = tmpRaw[:indexDdEnd]
tmpRaw = tmpRaw[indexDdEnd + 1 :]
elif re.search(r"^<table", tmpRaw) != None:
import rtfTable
table = rtfTable.rtfTable()
indexTableEnd = tmpRaw.index("</table>") + len("</table>")
tmpTableRaw = tmpRaw[:indexTableEnd]
tmpRaw = tmpRaw[indexTableEnd + 1 :]
# Set raw content
table.set_raw(tmpTableRaw)
# Update paragraph list
tmp_objs.append(table)
# Update content
tmp_content = tmp_content + table.get_content() + "\n"
elif re.search(r"^<image", tmpRaw) != None:
# import rtfImage
# table = rtfTable.rtfTable()
indexImageEnd = tmpRaw.index("</image>") + len("</image>")
# tmpTableRaw = tmpRaw[:indexTableEnd]
tmpRaw = tmpRaw[indexImageEnd + 1 :]
# Set raw content
# table.set_raw(tmpTableRaw)
# Update paragraph list
# tmp_objs.append(table)
# Update content
# tmp_content = tmp_content + table.get_content() + "\n"
elif re.search(r"^<frame", tmpRaw) != None:
import rtfFrame
frame = rtfFrame.rtfFrame()
indexFrameEnd = tmpRaw.index("</frame>") + len("</frame>")
tmpFrameRaw = tmpRaw[:indexFrameEnd]
tmpRaw = tmpRaw[indexFrameEnd + 1 :]
# Set raw content
frame.set_raw(tmpFrameRaw)
# Update paragraph list
tmp_objs.append(frame)
# Update content
tmp_content = tmp_content + frame.get_content() + "\n"
else:
break
self.members["_bottom"] = tmp_bottom
self.members["_top"] = tmp_top
self.members["_left"] = tmp_left
self.members["_right"] = tmp_right
self.members["_objs"] = tmp_objs
self.members["_content"] = tmp_content
